SENTIMENT ANALYSIS IN NEWS MEDIA HEADLINES IN 2019 PRESIDENTIAL ELECTIONS: EXPLORATORY RELIABILITY STUDY ANALYSIS.

EMILIO RECART; TOMAS ALVES SALGUIERO; FRANCO AGUSTIN BERNAL; DAMIAN FURMAN; JUAN MANUEL PEREZ; P. N. FERNÁNDEZ LARROSA

Leipzig

Simposio; 9th MindBrainBody Symposium 2022. Max Planck Institute; 2022

Max Planck Institute for Human Cognitive and Brain Sciences

In recent years, social networksoffer the facility to access political headnews instantly. Despite the growinginformation, there has been a spread of fake news generating a biasedperception to the politics. According to this, new computational models havebeen developed to identify and predict subjective perception towards socialpolitic content. Reliability has been a widely used tool in measuringagreements that different people arrive at the presentation of the samestimulus. By the other hand, sentimental analysis is a useful tool foranalyzing electoral behavior. The purpose of this study was to evaluate thesubjective perception of individuals for each presidential formula/force ofprincipal Argentina´s newspapers during 2019 elections using reliabilitycoefficient agreement. For this, 3 participants were recruited to classify 2257headlines of the principal country´s newspapers as positive, neutral, ornegative according to their perception. To minimize ideological bias, eachformula/force was replaced by a "Target". Krippendorf nominalreliability alpha metric yielded adequate inter agreements between theparticipants. With this tool, we found that Alberto Fernandez (the winningcandidate) was mentioned with positive connotation in 387,66 headlines (negative:239,66), while Mauricio Macri (the outgoing president and the second candidate)in 320,66 ones (negative: 430,66). Other candidates did not exceed 140 positivementions(negative: <16,66). According to this, using these metrics could bea useful tool for future studies for classifying the valence of the headlines.